In the past decade, there has been a surge of renewed interest in the study of developmental evolution. One approach that has been taken is to examine the expression patterns of a candidate gene in divergent taxa and to use these results to infer which aspects of a particular genetic pathway are either conserved or altered. Here we consider this approach from the perspective of the neo-Darwinian paradigm for evolutionary change. If adaptations are typically composed of large numbers of gene substitutions that are of small effect individually, then the candidate gene approach is unlikely to bridge the gap between developmental pattern and evolutionary process: changes in gene expression patterns may identify the steps in developmental pathways that have been altered during evolution but fail to identify the actual genetic changes that have occurred. On the other hand, there is growing support for the view that adaptations often involve large-effect genes; fortunately, the candidate gene approach is well suited to this type of genetic architecture.